[PATCH v2 0/2] regulator: support the HiSilicon STB PMOC voltage rails

[PATCH v2 0/2] regulator: support the HiSilicon STB PMOC voltage rails

[David Yang]

HiSilicon STB SoCs regulate their CPU and core supplies with PWM outputs of
the PMOC block feeding an external converter on the board. This adds the
binding for those registers and a driver for them.

v2:
 - naming and doc changes
v1: https://lore.kernel.org/r/20261008155727.3801095-1-mmyangfl@gmail.com

David Yang (2):
  regulator: dt-bindings: add hi3798cv200 regulator
  regulator: histb: add voltage regulator driver

 .../hisilicon,hi3798cv200-regulator.yaml      |  48 ++++++
 MAINTAINERS                                   |   7 +
 drivers/regulator/Kconfig                     |   7 +
 drivers/regulator/Makefile                    |   1 +
 drivers/regulator/histb-regulator.c           | 162 ++++++++++++++++++
 5 files changed, 225 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/regulator/hisilicon,hi3798cv200-regulator.yaml
 create mode 100644 drivers/regulator/histb-regulator.c

-- 
2.53.0

[PATCH v2 1/2] regulator: dt-bindings: add hi3798cv200 regulator

[David Yang]

The PMOC block of the HiSilicon STB SoCs (Hi3798cv200/mv100/mv200)
drives an external converter with a PWM signal. Add dt-binding doc.

Signed-off-by: David Yang <mmyangfl@gmail.com>
---
 .../hisilicon,hi3798cv200-regulator.yaml      | 48 +++++++++++++++++++
 1 file changed, 48 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/regulator/hisilicon,hi3798cv200-regulator.yaml

diff --git a/Documentation/devicetree/bindings/regulator/hisilicon,hi3798cv200-regulator.yaml b/Documentation/devicetree/bindings/regulator/hisilicon,hi3798cv200-regulator.yaml
new file mode 100644
index 000000000000..812d94aca1f5
--- /dev/null
+++ b/Documentation/devicetree/bindings/regulator/hisilicon,hi3798cv200-regulator.yaml
@@ -0,0 +1,48 @@
+# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/regulator/hisilicon,hi3798cv200-regulator.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: HiSilicon STB voltage regulator
+
+maintainers:
+  - David Yang <mmyangfl@gmail.com>
+
+description: |
+  HiSilicon STB SoCs (Hi3798cv200/mv100/mv200) provide PWM outputs in the PMOC
+  block which drive an external converter.
+
+  The voltage range depends on the board design: reference designs specify
+  0.65 V to 1.15 V for the Hi3798cv200, and 0.7 V to 1.25 V for the Hi3798mv200;
+  a customized board has to figure out its own. The regulator itself has no
+  other source of calibration.
+
+properties:
+  compatible:
+    const: hisilicon,hi3798cv200-regulator
+
+  reg:
+    maxItems: 1
+
+required:
+  - compatible
+  - reg
+  - regulator-min-microvolt
+  - regulator-max-microvolt
+
+allOf:
+  - $ref: regulator.yaml#
+
+unevaluatedProperties: false
+
+examples:
+  - |
+    vddcpu: regulator@8a23018 {
+        compatible = "hisilicon,hi3798cv200-regulator";
+        reg = <0x8a23018 0x4>;
+        regulator-name = "vdd-cpu";
+        regulator-min-microvolt = <650000>;
+        regulator-max-microvolt = <1150000>;
+        regulator-always-on;
+    };
-- 
2.53.0

Re: [PATCH v2 1/2] regulator: dt-bindings: add hi3798cv200 regulator

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] dt-bindings: regulator: missing hardware calibration properties causes overvoltage
- [Medium] dt-bindings: compatible restricts to single SoC despite supporting multiple variants
--

commit e41927e1676c51346b08990a591c57b883a2a88d
Author: David Yang <mmyangfl@gmail.com>

regulator: dt-bindings: add hi3798cv200 regulator

The PMOC block of the HiSilicon STB SoCs (Hi3798cv200/mv100/mv200)
drives an external converter with a PWM signal. Add dt-binding doc.

> diff --git a/Documentation/devicetree/bindings/regulator/hisilicon,hi3798cv200-regulator.yaml b/Documentation/devicetree/bindings/regulator/hisilicon,hi3798cv200-regulator.yaml
> new file mode 100644
> index 0000000000000..812d94aca1f59
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/regulator/hisilicon,hi3798cv200-regulator.yaml

[ ... ]

> +description: |
> +  HiSilicon STB SoCs (Hi3798cv200/mv100/mv200) provide PWM outputs in the PMOC
> +  block which drive an external converter.
> +
> +  The voltage range depends on the board design: reference designs specify
> +  0.65 V to 1.15 V for the Hi3798cv200, and 0.7 V to 1.25 V for the Hi3798mv200;
> +  a customized board has to figure out its own. The regulator itself has no
> +  other source of calibration.
> +
> +properties:
> +  compatible:
> +    const: hisilicon,hi3798cv200-regulator

[Severity: Medium]
Will this strict constraint reject device trees for the other supported SoCs?

The description explicitly mentions support for multiple HiSilicon STB SoCs
(Hi3798cv200/mv100/mv200) which have different voltage reference designs.
However, restricting the compatible property to a single exact value
(hisilicon,hi3798cv200-regulator) rejects device trees that use SoC-specific
compatible strings (e.g., hisilicon,hi3798mv200-regulator) or fallback lists.

This prevents the proper representation of the other supported hardware variants
without failing schema validation (dt_binding_check).

> +
> +  reg:
> +    maxItems: 1
> +
> +required:
> +  - compatible
> +  - reg
> +  - regulator-min-microvolt
> +  - regulator-max-microvolt
> +
> +allOf:
> +  - $ref: regulator.yaml#
> +
> +unevaluatedProperties: false

[Severity: High]
Does the lack of hardware calibration properties here cause overvoltage
in the driver?

By not including properties to describe the physical hardware bounds (e.g.,
duty-cycle mappings or pwm-regulator integration), strict validation prevents
defining these board-specific calibration values in the device tree.

As a result, the follow-up commit "regulator: histb: add voltage regulator
driver" relies on the standard OS policy constraints (regulator-min-microvolt
and regulator-max-microvolt) as physical 100% and 0% duty cycle calibration
boundaries:

drivers/regulator/histb-regulator.c:histb_regulator_probe() {
    ...
	priv->min_uV = init_data->constraints.min_uV;
	priv->max_uV = init_data->constraints.max_uV;
    ...
}

If a device tree author specifies a tighter policy constraint to protect
hardware (e.g., a maximum of 1.0V when the physical maximum is 1.15V),
a legitimate request for 1.0V outputs a 0% duty cycle:

drivers/regulator/histb-regulator.c:histb_regulator_set_voltage() {
    ...
	if (min_uV >= priv->max_uV)
		duty = 0;
    ...
}

This configures the hardware to physically deliver the unconstrained maximum
of 1.15V, completely bypassing the intended safety bounds and potentially
damaging connected components.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261009214404.722628-1-mmyangfl@gmail.com?part=1

Re: [PATCH v2 1/2] regulator: dt-bindings: add hi3798cv200 regulator

[David Yang]

On Sat, Oct 10, 2026 at 5:52 AM <sashiko-bot@kernel.org> wrote:

> [Severity: Medium]
> Will this strict constraint reject device trees for the other supported SoCs?
>
> The description explicitly mentions support for multiple HiSilicon STB SoCs
> (Hi3798cv200/mv100/mv200) which have different voltage reference designs.
> However, restricting the compatible property to a single exact value
> (hisilicon,hi3798cv200-regulator) rejects device trees that use SoC-specific
> compatible strings (e.g., hisilicon,hi3798mv200-regulator) or fallback lists.
>
> This prevents the proper representation of the other supported hardware variants
> without failing schema validation (dt_binding_check).

No intend to introduce more compatibles; follows
hisilicon,hi3798cv200-pcie conventions.

> [Severity: High]
> Does the lack of hardware calibration properties here cause overvoltage
> in the driver?
>
> By not including properties to describe the physical hardware bounds (e.g.,
> duty-cycle mappings or pwm-regulator integration), strict validation prevents
> defining these board-specific calibration values in the device tree.
>
> As a result, the follow-up commit "regulator: histb: add voltage regulator
> driver" relies on the standard OS policy constraints (regulator-min-microvolt
> and regulator-max-microvolt) as physical 100% and 0% duty cycle calibration
> boundaries:
>
> drivers/regulator/histb-regulator.c:histb_regulator_probe() {
>     ...
>         priv->min_uV = init_data->constraints.min_uV;
>         priv->max_uV = init_data->constraints.max_uV;
>     ...
> }
>
> If a device tree author specifies a tighter policy constraint to protect
> hardware (e.g., a maximum of 1.0V when the physical maximum is 1.15V),
> a legitimate request for 1.0V outputs a 0% duty cycle:
>
> drivers/regulator/histb-regulator.c:histb_regulator_set_voltage() {
>     ...
>         if (min_uV >= priv->max_uV)
>                 duty = 0;
>     ...
> }
>
> This configures the hardware to physically deliver the unconstrained maximum
> of 1.15V, completely bypassing the intended safety bounds and potentially
> damaging connected components.

pwm-regulator.yaml: This mode uses the regulator's maximum and minimum
supplied voltages specified in the regulator-{min,max}-microvolt
properties to calculate appropriate duty-cycle values.

Re: [PATCH v2 1/2] regulator: dt-bindings: add hi3798cv200 regulator

[Krzysztof Kozlowski]

On Sat, Oct 10, 2026 at 05:43:39AM +0800, David Yang wrote:
> The PMOC block of the HiSilicon STB SoCs (Hi3798cv200/mv100/mv200)
> drives an external converter with a PWM signal. Add dt-binding doc.
> 
> Signed-off-by: David Yang <mmyangfl@gmail.com>
> ---
>  .../hisilicon,hi3798cv200-regulator.yaml      | 48 +++++++++++++++++++
>  1 file changed, 48 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/regulator/hisilicon,hi3798cv200-regulator.yaml
> 
> diff --git a/Documentation/devicetree/bindings/regulator/hisilicon,hi3798cv200-regulator.yaml b/Documentation/devicetree/bindings/regulator/hisilicon,hi3798cv200-regulator.yaml
> new file mode 100644
> index 000000000000..812d94aca1f5
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/regulator/hisilicon,hi3798cv200-regulator.yaml
> @@ -0,0 +1,48 @@
> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
> +%YAML 1.2
> +---
> +$id: http://devicetree.org/schemas/regulator/hisilicon,hi3798cv200-regulator.yaml#
> +$schema: http://devicetree.org/meta-schemas/core.yaml#
> +
> +title: HiSilicon STB voltage regulator
> +
> +maintainers:
> +  - David Yang <mmyangfl@gmail.com>
> +
> +description: |
> +  HiSilicon STB SoCs (Hi3798cv200/mv100/mv200) provide PWM outputs in the PMOC
> +  block which drive an external converter.

SoCs rarely provide supplies for themselves (they are rather called
power domains in such case) and above wording suggests this is actually
something external. It also speaks about PWM but binding has no PWM.

> +
> +  The voltage range depends on the board design: reference designs specify
> +  0.65 V to 1.15 V for the Hi3798cv200, and 0.7 V to 1.25 V for the Hi3798mv200;
> +  a customized board has to figure out its own. The regulator itself has no
> +  other source of calibration.
> +
> +properties:
> +  compatible:
> +    const: hisilicon,hi3798cv200-regulator
> +
> +  reg:
> +    maxItems: 1
> +
> +required:
> +  - compatible
> +  - reg
> +  - regulator-min-microvolt
> +  - regulator-max-microvolt
> +
> +allOf:
> +  - $ref: regulator.yaml#
> +
> +unevaluatedProperties: false
> +
> +examples:
> +  - |
> +    vddcpu: regulator@8a23018 {

Drop unused label

> +        compatible = "hisilicon,hi3798cv200-regulator";
> +        reg = <0x8a23018 0x4>;

And one word is rarely a device, which adds to the confusion.

Best regards,
Krzysztof

Re: [PATCH v2 1/2] regulator: dt-bindings: add hi3798cv200 regulator

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 771 bytes --]

On Mon, Oct 12, 2026 at 11:11:33AM +0200, Krzysztof Kozlowski wrote:
> On Sat, Oct 10, 2026 at 05:43:39AM +0800, David Yang wrote:

> > +description: |
> > +  HiSilicon STB SoCs (Hi3798cv200/mv100/mv200) provide PWM outputs in the PMOC
> > +  block which drive an external converter.

> SoCs rarely provide supplies for themselves (they are rather called
> power domains in such case) and above wording suggests this is actually
> something external. It also speaks about PWM but binding has no PWM.

I think this is talking about something like pwm-regulator.c.

> > +        compatible = "hisilicon,hi3798cv200-regulator";
> > +        reg = <0x8a23018 0x4>;

> And one word is rarely a device, which adds to the confusion.

It is possible for a simple analog block...

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

[PATCH v2 2/2] regulator: histb: add voltage regulator driver

[David Yang]

Add a driver for the voltage rails on HiSilicon STB SoCs which are
controlled by the PMOC block through a PWM signal, as described by the
hisilicon,hi3798cv200-regulator binding.

The name conventions follow those of pcie-histb.c.

Signed-off-by: David Yang <mmyangfl@gmail.com>
---
 MAINTAINERS                         |   7 ++
 drivers/regulator/Kconfig           |   7 ++
 drivers/regulator/Makefile          |   1 +
 drivers/regulator/histb-regulator.c | 162 ++++++++++++++++++++++++++++
 4 files changed, 177 insertions(+)
 create mode 100644 drivers/regulator/histb-regulator.c

diff --git a/MAINTAINERS b/MAINTAINERS
index 65e8a4b5c90b..18b86d69ad07 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -11910,6 +11910,13 @@ S:	Maintained
 F:	Documentation/devicetree/bindings/mfd/hisilicon,hi6421-spmi-pmic.yaml
 F:	drivers/mfd/hi6421-spmi-pmic.c
 
+HISILICON STB VOLTAGE REGULATOR DRIVER
+M:	David Yang <mmyangfl@gmail.com>
+L:	linux-kernel@vger.kernel.org
+S:	Maintained
+F:	Documentation/devicetree/bindings/regulator/hisilicon,hi3798cv200-regulator.yaml
+F:	drivers/regulator/histb-regulator.c
+
 HISILICON TRUE RANDOM NUMBER GENERATOR V2 SUPPORT
 M:	Weili Qian <qianweili@huawei.com>
 S:	Maintained
diff --git a/drivers/regulator/Kconfig b/drivers/regulator/Kconfig
index 89789ac7a786..a059c5b2bafc 100644
--- a/drivers/regulator/Kconfig
+++ b/drivers/regulator/Kconfig
@@ -487,6 +487,13 @@ config REGULATOR_HI6421V600
 	  HiSilicon Hi6421v600 PMU / Codec IC.
 	  This is used on Kirin 3670 boards, like HiKey 970.
 
+config REGULATOR_HISI_STB
+	tristate "HiSilicon STB voltage regulator support"
+	depends on ARCH_HISI || COMPILE_TEST
+	help
+	  This driver provides support for the voltage regulators of the
+	  HiSilicon STB SoCs, including Hi3798cv200, Hi3798mv100, Hi3798mv200.
+
 config REGULATOR_ISL9305
 	tristate "Intersil ISL9305 regulator"
 	depends on I2C
diff --git a/drivers/regulator/Makefile b/drivers/regulator/Makefile
index 5a764cec8df8..a2efd2bc3e0d 100644
--- a/drivers/regulator/Makefile
+++ b/drivers/regulator/Makefile
@@ -57,6 +57,7 @@ obj-$(CONFIG_REGULATOR_HI6421) += hi6421-regulator.o
 obj-$(CONFIG_REGULATOR_HI6421V530) += hi6421v530-regulator.o
 obj-$(CONFIG_REGULATOR_HI6421V600) += hi6421v600-regulator.o
 obj-$(CONFIG_REGULATOR_HI655X) += hi655x-regulator.o
+obj-$(CONFIG_REGULATOR_HISI_STB) += histb-regulator.o
 obj-$(CONFIG_REGULATOR_ISL6271A) += isl6271a-regulator.o
 obj-$(CONFIG_REGULATOR_ISL9305) += isl9305.o
 obj-$(CONFIG_REGULATOR_FP9931) += fp9931.o
diff --git a/drivers/regulator/histb-regulator.c b/drivers/regulator/histb-regulator.c
new file mode 100644
index 000000000000..c6d8defc81da
--- /dev/null
+++ b/drivers/regulator/histb-regulator.c
@@ -0,0 +1,162 @@
+// SPDX-License-Identifier: MIT OR GPL-2.0-or-later
+/*
+ * Regulator driver for the PWM controlled voltage regulators of HiSilicon STB
+ * SoCs
+ *
+ * May work on Hi3716mv410/420 Hi3796mv100/200 Hi3798cv200/mv100/mv200 but not
+ * thoroughly tested
+ *
+ * Copyright (c) 2026 David Yang
+ */
+
+#include <linux/bitfield.h>
+#include <linux/err.h>
+#include <linux/io.h>
+#include <linux/math64.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/regulator/driver.h>
+#include <linux/regulator/machine.h>
+#include <linux/regulator/of_regulator.h>
+
+/*
+ * Duty cycle + 1 and period + 1 are counted in periods of the oscillator
+ * clock (usually 24 MHz), and a larger duty cycle means a lower output
+ * voltage.
+ */
+#define HISTB_DUTY		GENMASK(31, 16)
+#define HISTB_PERIOD		GENMASK(15, 0)
+
+/* 24 MHz / 240 = 100 kHz, arbitrarily chosen but should be enough */
+#define HISTB_PWM_PERIOD	240
+
+struct histb_regulator {
+	struct regulator_desc desc;
+	void __iomem *base;
+	int min_uV;
+	int max_uV;
+};
+
+static int histb_regulator_get_voltage(struct regulator_dev *rdev)
+{
+	struct histb_regulator *priv = rdev_get_drvdata(rdev);
+	u32 val;
+	u32 period;
+	u32 duty;
+	u64 uV;
+
+	val = readl_relaxed(priv->base);
+	period = FIELD_GET(HISTB_PERIOD, val);
+	duty = FIELD_GET(HISTB_DUTY, val);
+
+	/* an empty counter range leaves the output high */
+	if (duty <= 1 || period <= 1)
+		return priv->max_uV;
+
+	duty--;
+	period--;
+	if (duty >= period)
+		return priv->min_uV;
+
+	uV = (u64)(priv->max_uV - priv->min_uV) * duty;
+
+	return priv->max_uV - DIV_ROUND_CLOSEST_ULL(uV, period);
+}
+
+static int
+histb_regulator_set_voltage(struct regulator_dev *rdev, int min_uV, int max_uV,
+			    unsigned int *selector)
+{
+	struct histb_regulator *priv = rdev_get_drvdata(rdev);
+	unsigned int duty;
+	u32 val;
+
+	if (min_uV >= priv->max_uV)
+		duty = 0;
+	else if (min_uV <= priv->min_uV)
+		duty = HISTB_PWM_PERIOD;
+	else
+		/* rounding down keeps the output at or above the request */
+		duty = DIV_ROUND_DOWN_ULL((u64)(priv->max_uV - min_uV) *
+					  HISTB_PWM_PERIOD,
+					  priv->max_uV - priv->min_uV);
+
+	val = FIELD_PREP(HISTB_DUTY, duty + 1) |
+	      FIELD_PREP(HISTB_PERIOD, HISTB_PWM_PERIOD + 1);
+	writel_relaxed(val, priv->base);
+
+	return 0;
+}
+
+static const struct regulator_ops histb_regulator_ops = {
+	.get_voltage = histb_regulator_get_voltage,
+	.set_voltage = histb_regulator_set_voltage,
+};
+
+static int histb_regulator_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct histb_regulator *priv;
+	struct regulator_init_data *init_data;
+	struct regulator_config config = {};
+	struct regulator_dev *rdev;
+
+	priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
+	if (!priv)
+		return -ENOMEM;
+
+	init_data = of_get_regulator_init_data(dev, dev->of_node, &priv->desc);
+	if (!init_data)
+		return -ENOMEM;
+
+	priv->min_uV = init_data->constraints.min_uV;
+	priv->max_uV = init_data->constraints.max_uV;
+	if (priv->min_uV <= 0 || priv->max_uV <= priv->min_uV) {
+		dev_err(dev, "invalid voltage range (%d..%d uV)\n",
+			priv->min_uV, priv->max_uV);
+		return -EINVAL;
+	}
+
+	priv->base = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(priv->base))
+		return PTR_ERR(priv->base);
+
+	priv->desc.name = dev_name(dev);
+	priv->desc.type = REGULATOR_VOLTAGE;
+	priv->desc.ops = &histb_regulator_ops;
+	priv->desc.owner = THIS_MODULE;
+	priv->desc.continuous_voltage_range = true;
+
+	config.dev = dev;
+	config.init_data = init_data;
+	config.driver_data = priv;
+	config.of_node = dev->of_node;
+
+	rdev = devm_regulator_register(dev, &priv->desc, &config);
+	if (IS_ERR(rdev))
+		return dev_err_probe(dev, PTR_ERR(rdev),
+				     "failed to register %s\n",
+				     priv->desc.name);
+
+	return 0;
+}
+
+static const struct of_device_id histb_regulator_of_match[] = {
+	{ .compatible = "hisilicon,hi3798cv200-regulator" },
+	{}
+};
+MODULE_DEVICE_TABLE(of, histb_regulator_of_match);
+
+static struct platform_driver histb_regulator_driver = {
+	.probe = histb_regulator_probe,
+	.driver = {
+		.name = "hi3798cv200-regulator",
+		.of_match_table = histb_regulator_of_match,
+	},
+};
+module_platform_driver(histb_regulator_driver);
+
+MODULE_DESCRIPTION("HiSilicon STB regulator driver");
+MODULE_AUTHOR("David Yang <mmyangfl@gmail.com>");
+MODULE_LICENSE("Dual MIT/GPL");
-- 
2.53.0